EP0694906A1 - Method and system for speech recognition - Google Patents
Method and system for speech recognition Download PDFInfo
- Publication number
- EP0694906A1 EP0694906A1 EP95111784A EP95111784A EP0694906A1 EP 0694906 A1 EP0694906 A1 EP 0694906A1 EP 95111784 A EP95111784 A EP 95111784A EP 95111784 A EP95111784 A EP 95111784A EP 0694906 A1 EP0694906 A1 EP 0694906A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- speech
- noise
- feature vector
- frame
- feature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 239000013598 vector Substances 0.000 claims abstract description 164
- 238000010606 normalization Methods 0.000 claims abstract description 25
- 230000006978 adaptation Effects 0.000 claims description 6
- 230000000694 effects Effects 0.000 abstract description 8
- 239000000284 extract Substances 0.000 abstract description 5
- 230000001419 dependent effect Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000007476 Maximum Likelihood Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/10—Speech classification or search using distance or distortion measures between unknown speech and reference templates
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/12—Speech classification or search using dynamic programming techniques, e.g. dynamic time warping [DTW]
Definitions
- This invention relates generally to speech recognition and, more particularly, to a method and system for improving speech recognition through front-end normalization of feature vectors.
- Speech recognition generally involves two phases. The first phase is known as training. During training, the system "learns" speech by inputting a large sample of speech and generating models of the speech. The second phase is known as recognition. During recognition, the system attempts to recognize input speech by comparing the speech to the models generated during training and finding an exact match or a best match. Most speech recognition systems have a front-end that extracts some features from the input speech in the form of feature vectors. These feature vectors are used to generate the models during training and are compared to the generated models during recognition.
- a particular word or sentence should always be recognized as that word or sentence, regardless of the acoustical environment in which the word or sentence is spoken.
- mean normalization One attempt to solve this problem is known as mean normalization.
- the input speech feature vector is normalized by computing the mean of all the feature vectors extracted from the entire speech and subtracting the mean from the input speech feature vector using the function: where ( t ) is the normalized input speech feature vector, x( t ) is the raw input speech feature vector, and n is the number of feature vectors extracted from the entire speech.
- SNR-dependent signal-to-noise-ratio-dependent
- the input speech feature vector is normalized by computing the instantaneous SNR of the input speech and subtracting a correction vector that depends on the SNR from the input speech feature vector using the function: where ( t ) is the normalized input speech feature vector, x( t ) is the raw input speech feature vector, and y( SNR ) is the correction vector.
- the correction vectors are precomputed and stored in a look-up table with the corresponding SNR's.
- Mean normalization allows the input speech feature vectors to be dynamically adjusted but is not very accurate because it only computes a single mean for all of the feature vectors extracted from the entire speech.
- SNR-dependent normalization is more accurate than mean normalization because it computes varying correction vectors depending on the SNR of the input speech but it does not dynamically update the values of the correction vectors. Therefore, a solution is needed that both is accurate and dynamically updates the values used to normalize the input speech feature vectors.
- One aspect of the present invention provides a method and system for improving speech recognition through front-end normalization of feature vectors.
- speech to be recognized is spoken into a microphone, amplified by an amplifier, and converted from an analog signal to a digital signal by an analog-to-digital ("A/D") converter.
- the digital signal from the A/D converter is input to a feature extractor that breaks down the signal into frames of speech and then extracts a feature vector from each of the frames.
- the feature vector is input to an input normalizer that normalizes the vector.
- the normalized feature vector is input to a pattern matcher that compares the normalized vector to feature models stored in a database to find an exact match or a best match.
- the input normalizer of the present invention normalizes the feature vector by computing a correction vector and subtracting the correction vector from the feature vector.
- the correction vector is computed based on the probability of the current frame of speech being noise and based on the average noise and speech feature vectors for a current utterance and the database of utterances.
- the normalization of feature vectors reduces the effect of changes in the acoustical environment on the feature vectors. By reducing the effect of changes in the acoustical environment on the feature vectors, the input normalizer of the present invention improves the accuracy of the speech recognition system.
- the preferred embodiment of the present invention provides a method and system for improving speech recognition through front-end normalization of feature vectors.
- the normalization of feature vectors reduces the effect of changes in the acoustical environment on the feature vectors. Such changes could result, for example, from changes in the microphone used, the background noise, the distance between the speaker's mouth and the microphone, and the room acoustics. Without normalization, the effect of changes in the acoustical environment on the feature vectors could cause the same speech to be recognized as different speech. This could occur because the acoustical environment affects the feature vectors extracted from speech. Thus, different feature vectors may be extracted from the same speech if spoken in different acoustical environments. By reducing the effect of changes in the acoustical environment on the feature vectors, the input normalizer of the present invention improves the accuracy of the speech recognition system.
- FIG. 1 illustrates a speech recognition system 10 incorporating the principles of the present invention.
- speech to be recognized is spoken into a microphone 12, amplified by an amplifier 14, and converted from an analog signal to a digital signal by an analog-to-digital (“A/D") converter 16.
- the microphone 12, amplifier 14, and A/D converter 16 are conventional components and are well-known in the art.
- the digital signal from the A/D converter 16 is input to a computer system 18. More specifically, the digital signal is input to a feature extractor 20 that extracts certain features from the signal in the form of feature vectors.
- Speech is composed of utterances. An utterance is the spoken realization of a sentence and typically represents 1 to 10 seconds of speech. Each utterance is broken down into evenly-spaced time intervals called frames.
- a frame typically represents 10 milliseconds of speech.
- a feature vector is extracted from each frame of speech. That is, the feature extractor 20 breaks down the digital signal from the A/D converter 16 into frames of speech and then extracts a feature vector from each of the frames.
- the feature vector extracted from each frame of speech comprises cepstral vectors. Cepstral vectors, and the methods used to extract cepstral vectors from speech, are well-known in the art.
- the feature vector is then input to an input normalizer 22 that normalizes the vector.
- the normalization of the feature vector reduces the effect of changes in the acoustical environment on the feature vector.
- the normalized feature vector is then input to a pattern matcher 24 that compares the normalized vector to feature models stored in a database 26 to find an exact match or a best match.
- the feature models stored in the database 26 were generated from known speech. If there is an acceptable match, the known speech corresponding to the matching feature model is output. Otherwise, a message indicating that the speech could not be recognized is output.
- Typical pattern matchers are based on networks trained by statistical methods, such as hidden Markov models or neural networks. However, other pattern matchers may be used. Such pattern matchers are well-known in the art.
- the input normalizer 22 receives the feature vector x j for the current frame j , where j is an index (step 210).
- the feature vector comprises cepstral vectors.
- a cepstral vector is a set of coefficients derived from the energy in different frequency bands by taking the Discrete Cosine Transform ("DCT") of the logarithm of such energies.
- the feature vector comprises a static cepstral vector augmented with its first and second order derivatives with time, the delta cepstral vector and the delta-delta cepstral vector, respectively.
- Each cepstral vector comprises a set of thirteen cepstral coefficients.
- cepstral vectors having a different number of cepstral coefficients may be used.
- other forms of feature vectors may be used.
- the input normalizer 22 computes a normalized feature vector j using the function (step 214): While the feature vector comprises the three cepstral vectors discussed above, in the preferred embodiment of the present invention, only the static cepstral vector is normalized, the delta cepstral vector and the delta-delta cepstral vector are not normalized.
- the computation of the correction vector r(x j ) is simplified based on certain assumptions and estimations.
- ⁇ is the a priori probability of the current frame j being noise
- N (x j , n j -1 , ⁇ n ( j -1) ) and N (x j , s j -1 , ⁇ s ( j -1) ) are the Gaussian probability density functions ("pdf's") for noise and speech, respectively
- ⁇ n ( j -1) and ⁇ s ( j -1) are the covariance matrices for noise and speech, respectively.
- the Gaussian pdf's for noise and speech, N (x j , n j -1 , ⁇ n ( j -1) ) and N ( x j , s j -1 , ⁇ s ( j -1) ), are represented using the standard function for Gaussian pdf's: and where q is the dimension of x j , exp is the exponential function, and T represents the transpose function.
- the a posteriori probability of the current frame j being noise p j is represented by the sigmoid function: where where d (x j ) or d j is the distortion.
- the distortion is an indication of whether a signal is noise or speech. If the distortion is largely negative, the signal is noise; if the distortion is largely positive, the signal is speech; if the distortion is zero, the signal may be noise or speech.
- EM Estimate-Maximize
- the EM algorithm is discussed in N.M. Laird, A.P. Dempster, and D.B. Rubin, "Maximum Likelihood from Incomplete Data via the EM Algorithm," Annals Royal Statistical Society , 1-38, December 1967.
- the EM algorithm generates maximum likelihood estimates of the values by refining previous estimates based on new values.
- This algorithm uses a window function over which the estimates are refined.
- the window function defines the interval of time over which past estimates are used to refine the current estimates.
- the standard EM algorithm uses a rectangular window function.
- a rectangular window function gives equal weight to the data over the entire window.
- the modified version of the EM algorithm used in the preferred embodiment of the present invention uses an exponential window function.
- An exponential window function gives more weight to recent data in the window.
- the values of n , s , ⁇ n , ⁇ s , and ⁇ are estimated using the functions: where w k is the exponential window function.
- the rate of adaptation determines how much weight is given to past data relative to the current data. The smaller ⁇ is, the less weight that is given to past data relative to the current data; the larger ⁇ is, the more weight that is given to past data relative to the current data.
- the value of ⁇ is computed using the function: where T is a time constant and F s is the sampling frequency of the A/D converter 16.
- T is a time constant
- F s is the sampling frequency of the A/D converter 16.
- separate ⁇ 's are used for noise and for speech. The use of separate ⁇ 's allows noise and speech to be adapted at different rates.
- ⁇ n and ⁇ s are the parameters that control the rate of adaptation for noise and speech, respectively.
- step 310 values for ⁇ n and ⁇ s are selected.
- the values for ⁇ n and ⁇ s are selected based on the desired rate of adaptation (as discussed above).
- step 312 the value of j is set equal to zero (step 312) and initial values for n , s , ⁇ n , ⁇ s , and ⁇ are estimated (step 314).
- the initial values for n , s , ⁇ n , ⁇ s , and ⁇ are estimated from the database of utterances 26 using standard EM techniques.
- step 330 If frame j is not the last frame in the current utterance, j is incremented (step 330) and steps 316 through 326 are repeated for the next frame. If frame j is the last frame in the current utterance, the input normalizer 22 determines whether the current utterance is the last utterance (step 332). If the current utterance is not the last utterance, j is reset to zero (step 334), the values of n , s , ⁇ n , ⁇ s , and ⁇ are reset to the estimated initial values (step 336), and steps 316 through 326 are repeated for each frame in the next utterance. If the current utterance is the last utterance, the input normalizer 22 returns.
- the last term could be eliminated from the function (Eq. 38) used to compute the distortion d j .
- This term does not significantly affect the value of the distortion d j , but is expensive to compute because it involves a logarithm.
- the a posteriori probability of the current frame j being noise p j could be computed using a look-up table. This table would contain the possible values for the distortion d j and the corresponding values for the a posteriori probability p j .
- n , s , ⁇ n , and ⁇ s could be updated every few frames, instead of every frame, and the value of ⁇ could be kept at its initial value and not updated at all.
- each utterance in the database 26 is normalized according to the principles of the present invention and then the system is retrained using the database of normalized utterances. The database of normalized utterances is then used during recognition as described above.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
Description
- This invention relates generally to speech recognition and, more particularly, to a method and system for improving speech recognition through front-end normalization of feature vectors.
- A variety of speech recognition systems have been developed. These systems enable computers to understand speech. This ability is useful for inputting commands or data into computers. Speech recognition generally involves two phases. The first phase is known as training. During training, the system "learns" speech by inputting a large sample of speech and generating models of the speech. The second phase is known as recognition. During recognition, the system attempts to recognize input speech by comparing the speech to the models generated during training and finding an exact match or a best match. Most speech recognition systems have a front-end that extracts some features from the input speech in the form of feature vectors. These feature vectors are used to generate the models during training and are compared to the generated models during recognition.
- One problem with such speech recognition systems arises when there are changes in the acoustical environment during and between training and recognition. Such changes could result, for example, from changes in the microphone used, the background noise, the distance between the speaker's mouth and the microphone, and the room acoustics. If changes occur, the system may not work very well because the acoustical environment affects the feature vectors extracted from speech. Thus, different feature vectors may be extracted from the same speech if spoken in different acoustical environments. Since the acoustical environment will rarely remain constant, it is desirable for a speech recognition system to be robust to changes in the acoustical environment. A particular word or sentence should always be recognized as that word or sentence, regardless of the acoustical environment in which the word or sentence is spoken. Some attempts to solve the problem of changes in the acoustical environment have focused on normalizing the input speech feature vectors to reduce the effect of such changes.
- One attempt to solve this problem is known as mean normalization. Using mean normalization, the input speech feature vector is normalized by computing the mean of all the feature vectors extracted from the entire speech and subtracting the mean from the input speech feature vector using the function:
where (t) is the normalized input speech feature vector, x(t) is the raw input speech feature vector, and n is the number of feature vectors extracted from the entire speech. - Another attempt to solve this problem is known as signal-to-noise-ratio-dependent ("SNR-dependent") normalization. Using SNR-dependent normalization, the input speech feature vector is normalized by computing the instantaneous SNR of the input speech and subtracting a correction vector that depends on the SNR from the input speech feature vector using the function:
where (t) is the normalized input speech feature vector, x(t) is the raw input speech feature vector, and y(SNR) is the correction vector. The correction vectors are precomputed and stored in a look-up table with the corresponding SNR's. - None of the prior attempts to solve the problem of changes in the acoustical environment during and between training and recognition have been very successful. Mean normalization allows the input speech feature vectors to be dynamically adjusted but is not very accurate because it only computes a single mean for all of the feature vectors extracted from the entire speech. SNR-dependent normalization is more accurate than mean normalization because it computes varying correction vectors depending on the SNR of the input speech but it does not dynamically update the values of the correction vectors. Therefore, a solution is needed that both is accurate and dynamically updates the values used to normalize the input speech feature vectors.
- One aspect of the present invention provides a method and system for improving speech recognition through front-end normalization of feature vectors. In a speech recognition system of the present invention, speech to be recognized is spoken into a microphone, amplified by an amplifier, and converted from an analog signal to a digital signal by an analog-to-digital ("A/D") converter. The digital signal from the A/D converter is input to a feature extractor that breaks down the signal into frames of speech and then extracts a feature vector from each of the frames. The feature vector is input to an input normalizer that normalizes the vector. The normalized feature vector is input to a pattern matcher that compares the normalized vector to feature models stored in a database to find an exact match or a best match.
- The input normalizer of the present invention normalizes the feature vector by computing a correction vector and subtracting the correction vector from the feature vector. The correction vector is computed based on the probability of the current frame of speech being noise and based on the average noise and speech feature vectors for a current utterance and the database of utterances. The normalization of feature vectors reduces the effect of changes in the acoustical environment on the feature vectors. By reducing the effect of changes in the acoustical environment on the feature vectors, the input normalizer of the present invention improves the accuracy of the speech recognition system.
-
- Figure 1 is a block diagram illustrating a speech recognition system incorporating the principles of the present invention;
- Figure 2 is a high level flow chart illustrating the steps performed by an input normalizer of the system of Figure 1; and
- Figures 3A and 3B collectively are a high level flow chart illustrating the steps performed in the normalization of feature vectors in the system of Figure 1.
- The preferred embodiment of the present invention provides a method and system for improving speech recognition through front-end normalization of feature vectors. The normalization of feature vectors reduces the effect of changes in the acoustical environment on the feature vectors. Such changes could result, for example, from changes in the microphone used, the background noise, the distance between the speaker's mouth and the microphone, and the room acoustics. Without normalization, the effect of changes in the acoustical environment on the feature vectors could cause the same speech to be recognized as different speech. This could occur because the acoustical environment affects the feature vectors extracted from speech. Thus, different feature vectors may be extracted from the same speech if spoken in different acoustical environments. By reducing the effect of changes in the acoustical environment on the feature vectors, the input normalizer of the present invention improves the accuracy of the speech recognition system.
- Figure 1 illustrates a
speech recognition system 10 incorporating the principles of the present invention. In this system, speech to be recognized is spoken into amicrophone 12, amplified by anamplifier 14, and converted from an analog signal to a digital signal by an analog-to-digital ("A/D")converter 16. Themicrophone 12,amplifier 14, and A/D converter 16 are conventional components and are well-known in the art. The digital signal from the A/D converter 16 is input to acomputer system 18. More specifically, the digital signal is input to afeature extractor 20 that extracts certain features from the signal in the form of feature vectors. Speech is composed of utterances. An utterance is the spoken realization of a sentence and typically represents 1 to 10 seconds of speech. Each utterance is broken down into evenly-spaced time intervals called frames. A frame typically represents 10 milliseconds of speech. A feature vector is extracted from each frame of speech. That is, thefeature extractor 20 breaks down the digital signal from the A/D converter 16 into frames of speech and then extracts a feature vector from each of the frames. In the preferred embodiment of the present invention, the feature vector extracted from each frame of speech comprises cepstral vectors. Cepstral vectors, and the methods used to extract cepstral vectors from speech, are well-known in the art. - The feature vector is then input to an
input normalizer 22 that normalizes the vector. The normalization of the feature vector reduces the effect of changes in the acoustical environment on the feature vector. The normalized feature vector is then input to apattern matcher 24 that compares the normalized vector to feature models stored in adatabase 26 to find an exact match or a best match. The feature models stored in thedatabase 26 were generated from known speech. If there is an acceptable match, the known speech corresponding to the matching feature model is output. Otherwise, a message indicating that the speech could not be recognized is output. Typical pattern matchers are based on networks trained by statistical methods, such as hidden Markov models or neural networks. However, other pattern matchers may be used. Such pattern matchers are well-known in the art. - The steps performed by the
input normalizer 22 are shown in Figure 2. Theinput normalizer 22 receives the feature vector x j for the current frame j, where j is an index (step 210). In the preferred embodiment of the present invention, the feature vector comprises cepstral vectors. A cepstral vector is a set of coefficients derived from the energy in different frequency bands by taking the Discrete Cosine Transform ("DCT") of the logarithm of such energies. In the preferred embodiment, the feature vector comprises a static cepstral vector augmented with its first and second order derivatives with time, the delta cepstral vector and the delta-delta cepstral vector, respectively. Each cepstral vector comprises a set of thirteen cepstral coefficients. However, one of ordinary skill in the art will appreciate that cepstral vectors having a different number of cepstral coefficients may be used. Additionally, one of ordinary skill in the art will appreciate that other forms of feature vectors may be used. - Next, the
input normalizer 22 computes a correction vector r(x j ) or r j using the function (step 212):utterances 26. The computation of n, s, n avg , and s avg will be discussed below. Lastly, theinput normalizer 22 computes a normalized feature vector j using the function (step 214):
While the feature vector comprises the three cepstral vectors discussed above, in the preferred embodiment of the present invention, only the static cepstral vector is normalized, the delta cepstral vector and the delta-delta cepstral vector are not normalized. - The computation of the correction vector r(x j ) is simplified based on certain assumptions and estimations. First, assume that noise and speech follow a Gaussian distribution. Based on this assumption, the a posteriori probability of the current frame j being noise p j is computed using the function:
where ξ is the a priori probability of the current frame j being noise, N (x j ,n j -1,Σn (j-1) ) and N (x j ,s j -1,Σs (j-1) ) are the Gaussian probability density functions ("pdf's") for noise and speech, respectively, and Σn (j-1) and Σs (j-1) are the covariance matrices for noise and speech, respectively. The Gaussian pdf's for noise and speech, N (x j ,n j -1,Σn (j-1) ) and N (x j ,s j -1,Σs(j-1)), are represented using the standard function for Gaussian pdf's:
and
where q is the dimension of x j , exp is the exponential function, and T represents the transpose function. - Then, the a posteriori probability of the current frame j being noise p j is represented by the sigmoid function:
where
where d (x j ) or d j is the distortion. The distortion is an indication of whether a signal is noise or speech. If the distortion is largely negative, the signal is noise; if the distortion is largely positive, the signal is speech; if the distortion is zero, the signal may be noise or speech. - Second, assume that the components of x j are independent of one another. Based on this assumption, Σn and Σs are modelled using diagonal covariance matrices σ n and σ s , respectively. Thus, d (x j ) is represented using the function:
where q is the dimension of σ n and σ s . Further, the most important factor in discriminating noise from speech is the power term (l=0). Thus, d (x j ) is approximated using the function:
Next, the values of n, s, σ n , σ s , and ξ are estimated using a modified version of the well-known Estimate-Maximize ("EM") algorithm. The EM algorithm is discussed in N.M. Laird, A.P. Dempster, and D.B. Rubin, "Maximum Likelihood from Incomplete Data via the EM Algorithm," Annals Royal Statistical Society, 1-38, December 1967. The EM algorithm generates maximum likelihood estimates of the values by refining previous estimates based on new values. This algorithm uses a window function over which the estimates are refined. The window function defines the interval of time over which past estimates are used to refine the current estimates. The standard EM algorithm uses a rectangular window function. A rectangular window function gives equal weight to the data over the entire window. The modified version of the EM algorithm used in the preferred embodiment of the present invention uses an exponential window function. An exponential window function gives more weight to recent data in the window. Thus, the values of n, s, σ n , σ s , and ξ are estimated using the functions:
where w k is the exponential window function. - The exponential window function w k is represented by:
where T is a time constant and F s is the sampling frequency of the A/D converter 16. In the preferred embodiment of the present invention, separate α's are used for noise and for speech. The use of separate α's allows noise and speech to be adapted at different rates. In the preferred embodiment in which separate α's are used, a smaller α is used for noise than for speech. Thus, the functions used to estimate the values of n, s, σ n , σ s , and ξ are reduced to:
where - The steps performed in the normalization of a feature vector are shown in Figures 3A and 3B. First, values for α n and α s are selected (step 310). The values for α n and α s are selected based on the desired rate of adaptation (as discussed above). Additionally, the value of j is set equal to zero (step 312) and initial values for n, s, σ n , σ s , and ξ are estimated (step 314). The initial values for n, s, σ n , σ s , and ξ are estimated from the database of
utterances 26 using standard EM techniques.
Then, the feature vector x j for the current frame j is received (step 316). The distortion d j is computed using the function (step 318):
The a posteriori probability of the current frame j being noise p j is computed using the function (step 320):
The correction vector r j is computed using the function (step 322):
for l=0,1,...,m
The values of n, s, σ n , σ s , and ξ are updated using the functions (step 326):
whereinput normalizer 22 determines whether frame j is the last frame in the current utterance (step 326). If frame j is not the last frame in the current utterance, j is incremented (step 330) and steps 316 through 326 are repeated for the next frame. If frame j is the last frame in the current utterance, theinput normalizer 22 determines whether the current utterance is the last utterance (step 332). If the current utterance is not the last utterance, j is reset to zero (step 334), the values of n, s, σ n , σ s , and ξ are reset to the estimated initial values (step 336), and steps 316 through 326 are repeated for each frame in the next utterance. If the current utterance is the last utterance, theinput normalizer 22 returns. - In order to reduce the computational complexity of the
input normalizer 22 of the present invention, one of ordinary skill in the art will appreciate that several modifications could be made to the input normalizer. First, the last term could be eliminated from the function (Eq. 38) used to compute the distortion d j . This term does not significantly affect the value of the distortion d j , but is expensive to compute because it involves a logarithm. Additionally, the a posteriori probability of the current frame j being noise p j could be computed using a look-up table. This table would contain the possible values for the distortion d j and the corresponding values for the a posteriori probability p j . Lastly, the values of n, s, σ n , and σ s could be updated every few frames, instead of every frame, and the value of ξ could be kept at its initial value and not updated at all. Each of these modifications will improve the efficiency of theinput normalizer 22 without significantly affecting the accuracy of the input normalizer. - While the invention has described the normalization of feature vectors only during recognition, the preferred embodiment of the present invention involves the normalization of feature vectors during training as well. Specifically, each utterance in the
database 26 is normalized according to the principles of the present invention and then the system is retrained using the database of normalized utterances. The database of normalized utterances is then used during recognition as described above. - One of ordinary skill in the art will now appreciate that the present invention provides a method and system for improving speech recognition through front-end normalization of feature vectors. Although the present invention has been shown and described with reference to a preferred embodiment, equivalent alterations and modifications will occur to those skilled in the art upon reading and understanding this specification. The present invention includes all such equivalent alterations and modifications and is limited only by the scope of the following claims.
Claims (21)
- A method for improving speech recognition through front-end normalization of feature vectors, speech comprising utterances, each utterance comprising frames of speech, each frame of speech being represented by a feature vector, the method comprising the steps of:
providing a database of known utterances, the database of utterances having an average noise feature vector and an average speech feature vector;
receiving a feature vector representing a frame of speech in an utterance to be recognized, the frame of speech having a probability of being noise, the utterance having an average noise feature vector and an average speech feature vector;
computing a correction vector based on the probability of the frame of speech being noise and based on the average noise and speech feature vectors for the utterance and the database of utterances; and
computing a normalized feature vector based on the feature vector and the correction vector. - The method of claim 1, wherein the step of receiving a feature vector comprises the step of receiving a cepstral vector.
- The method of claim 1, wherein the probability of the frame of speech being noise and the average noise and speech feature vectors for the utterance are updated for each frame of speech.
- The method of claim 1, wherein the step of computing a correction vector includes the steps of:
computing the probability of the frame of speech being noise based on a distortion measure of the frame of speech;
computing the average noise and speech feature vectors for the utterance;
computing the average noise and speech feature vectors for the database of utterances; and
computing the correction vector based on the probability of the frame of speech being noise and the differences between the average noise and speech feature vectors for the utterance and the database of utterances. - A method for improving speech recognition through front-end normalization of feature vectors, speech comprising utterances, each utterance comprising frames of speech, each frame of speech being represented by a feature vector, the method comprising the steps of:
providing a database of known utterances, the database of utterances having an average noise feature vector and an average speech feature vector; receiving a feature vector x j representing a
frame of speech j in an utterance to be recognized, the frame of speech having an a posteriori probability of being noise, the utterance having an average noise feature vector and an average speech feature vector;
computing a correction vector r(x j ) as:
computing a normalized feature vector j as: - The method of claim 5, wherein the step of receiving a feature vector includes the step of receiving a cepstral vector.
- The method of claim 5, wherein the a posteriori probability of the frame of speech being noise and the average noise and speech feature vectors for the utterance are updated for each frame of speech.
- The method of claim 5, wherein the a posteriori probability of the frame of speech j being noise p j is computed as:
- A system for improving speech recognition through front-end normalization of feature vectors, speech comprising utterances, each utterance comprising frames of speech, each frame of speech being represented by a feature vector, the system comprising:
a database of known utterances, the database of utterances having an average noise feature vector and an average speech feature vector; and
an input normalizer for:
receiving a feature vector representing a frame of speech in an utterance to be recognized, the frame of speech having a probability of being noise, the utterance having an average noise feature vector and an average speech feature vector;
computing a correction vector based on the probability of the frame of speech being noise and based on the average noise and speech feature vectors for the utterance and the database of utterances; and
computing a normalized feature vector based on the feature vector and the correction vector. - A system for improving speech recognition through front-end normalization of feature vectors, speech comprising utterances, each utterance comprising frames of speech, each frame of speech being represented by a feature vector, the system comprising:
a database of known utterances, the utterances being represented by feature models, the database of utterances having an average noise feature vector and an average speech feature vector;
a feature extractor for extracting a feature vector from a frame of speech in an utterance to be recognized, the frame of speech having a probability of being noise, the utterance having an average noise feature vector and an average speech feature vector;
an input normalizer for normalizing the feature vector by: (i) computing a correction vector based on the probability of the frame of speech being noise and based on the average noise and speech feature vectors for the utterance and the database of utterances, and (ii) computing a normalized feature vector based on the feature vector and the correction vector; and
a pattern matcher for comparing the normalized feature vector to the feature models in the database.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US283271 | 1994-07-29 | ||
US08/283,271 US5604839A (en) | 1994-07-29 | 1994-07-29 | Method and system for improving speech recognition through front-end normalization of feature vectors |
Publications (2)
Publication Number | Publication Date |
---|---|
EP0694906A1 true EP0694906A1 (en) | 1996-01-31 |
EP0694906B1 EP0694906B1 (en) | 2000-09-06 |
Family
ID=23085287
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP95111784A Expired - Lifetime EP0694906B1 (en) | 1994-07-29 | 1995-07-26 | Method and system for speech recognition |
Country Status (4)
Country | Link |
---|---|
US (1) | US5604839A (en) |
EP (1) | EP0694906B1 (en) |
JP (1) | JPH08110793A (en) |
DE (1) | DE69518705T2 (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0871157A2 (en) * | 1997-04-11 | 1998-10-14 | Nokia Mobile Phones Ltd. | A method and a device for recognising speech |
EP1113419A1 (en) * | 1999-12-28 | 2001-07-04 | Sony Corporation | Model adaptive apparatus and model adaptive method, recording medium, and pattern recognition apparatus |
EP1199712A2 (en) * | 2000-10-16 | 2002-04-24 | Microsoft Corporation | Noise reduction method |
EP1246165A1 (en) * | 2001-03-28 | 2002-10-02 | Matsushita Electric Industrial Co., Ltd. | Keyword detection in a noisy signal |
EP1326233A2 (en) * | 2001-12-28 | 2003-07-09 | Kabushiki Kaisha Toshiba | Apparatus and method for speech recognition in noise |
WO2005036525A1 (en) * | 2003-10-08 | 2005-04-21 | Philips Intellectual Property & Standards Gmbh | Adaptation of environment mismatch for speech recognition systems |
GB2422237A (en) * | 2004-12-21 | 2006-07-19 | Fluency Voice Technology Ltd | Dynamic coefficients determined from temporally adjacent speech frames |
US7117148B2 (en) | 2002-04-05 | 2006-10-03 | Microsoft Corporation | Method of noise reduction using correction vectors based on dynamic aspects of speech and noise normalization |
CN111489754A (en) * | 2019-01-28 | 2020-08-04 | 国家电网有限公司客户服务中心 | Telephone traffic data analysis method based on intelligent voice technology |
Families Citing this family (44)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2797949B2 (en) * | 1994-01-31 | 1998-09-17 | 日本電気株式会社 | Voice recognition device |
US5848246A (en) | 1996-07-01 | 1998-12-08 | Sun Microsystems, Inc. | Object-oriented system, method and article of manufacture for a client-server session manager in an interprise computing framework system |
US6304893B1 (en) | 1996-07-01 | 2001-10-16 | Sun Microsystems, Inc. | Object-oriented system, method and article of manufacture for a client-server event driven message framework in an interprise computing framework system |
US5987245A (en) | 1996-07-01 | 1999-11-16 | Sun Microsystems, Inc. | Object-oriented system, method and article of manufacture (#12) for a client-server state machine framework |
US6272555B1 (en) | 1996-07-01 | 2001-08-07 | Sun Microsystems, Inc. | Object-oriented system, method and article of manufacture for a client-server-centric interprise computing framework system |
US6266709B1 (en) | 1996-07-01 | 2001-07-24 | Sun Microsystems, Inc. | Object-oriented system, method and article of manufacture for a client-server failure reporting process |
US6424991B1 (en) | 1996-07-01 | 2002-07-23 | Sun Microsystems, Inc. | Object-oriented system, method and article of manufacture for a client-server communication framework |
US6434598B1 (en) | 1996-07-01 | 2002-08-13 | Sun Microsystems, Inc. | Object-oriented system, method and article of manufacture for a client-server graphical user interface (#9) framework in an interprise computing framework system |
US5999972A (en) | 1996-07-01 | 1999-12-07 | Sun Microsystems, Inc. | System, method and article of manufacture for a distributed computer system framework |
US6038590A (en) | 1996-07-01 | 2000-03-14 | Sun Microsystems, Inc. | Object-oriented system, method and article of manufacture for a client-server state machine in an interprise computing framework system |
JP3195752B2 (en) * | 1997-02-28 | 2001-08-06 | シャープ株式会社 | Search device |
JP3962445B2 (en) * | 1997-03-13 | 2007-08-22 | キヤノン株式会社 | Audio processing method and apparatus |
KR100450787B1 (en) * | 1997-06-18 | 2005-05-03 | 삼성전자주식회사 | Speech Feature Extraction Apparatus and Method by Dynamic Spectralization of Spectrum |
KR100442825B1 (en) * | 1997-07-11 | 2005-02-03 | 삼성전자주식회사 | Method for compensating environment for voice recognition, particularly regarding to improving performance of voice recognition system by compensating polluted voice spectrum closely to real voice spectrum |
US5946653A (en) * | 1997-10-01 | 1999-08-31 | Motorola, Inc. | Speaker independent speech recognition system and method |
US6173258B1 (en) * | 1998-09-09 | 2001-01-09 | Sony Corporation | Method for reducing noise distortions in a speech recognition system |
US6768979B1 (en) * | 1998-10-22 | 2004-07-27 | Sony Corporation | Apparatus and method for noise attenuation in a speech recognition system |
US6308155B1 (en) * | 1999-01-20 | 2001-10-23 | International Computer Science Institute | Feature extraction for automatic speech recognition |
GB2349259B (en) | 1999-04-23 | 2003-11-12 | Canon Kk | Speech processing apparatus and method |
US6804640B1 (en) * | 2000-02-29 | 2004-10-12 | Nuance Communications | Signal noise reduction using magnitude-domain spectral subtraction |
EP1229517B1 (en) * | 2001-02-06 | 2005-05-04 | Sony International (Europe) GmbH | Method for recognizing speech with noise-dependent variance normalization |
US6985858B2 (en) * | 2001-03-20 | 2006-01-10 | Microsoft Corporation | Method and apparatus for removing noise from feature vectors |
US6959276B2 (en) * | 2001-09-27 | 2005-10-25 | Microsoft Corporation | Including the category of environmental noise when processing speech signals |
US7139703B2 (en) * | 2002-04-05 | 2006-11-21 | Microsoft Corporation | Method of iterative noise estimation in a recursive framework |
US6944590B2 (en) | 2002-04-05 | 2005-09-13 | Microsoft Corporation | Method of iterative noise estimation in a recursive framework |
US7107210B2 (en) * | 2002-05-20 | 2006-09-12 | Microsoft Corporation | Method of noise reduction based on dynamic aspects of speech |
US7174292B2 (en) | 2002-05-20 | 2007-02-06 | Microsoft Corporation | Method of determining uncertainty associated with acoustic distortion-based noise reduction |
US7103540B2 (en) | 2002-05-20 | 2006-09-05 | Microsoft Corporation | Method of pattern recognition using noise reduction uncertainty |
US7047047B2 (en) | 2002-09-06 | 2006-05-16 | Microsoft Corporation | Non-linear observation model for removing noise from corrupted signals |
US7165026B2 (en) * | 2003-03-31 | 2007-01-16 | Microsoft Corporation | Method of noise estimation using incremental bayes learning |
US7720675B2 (en) * | 2003-10-27 | 2010-05-18 | Educational Testing Service | Method and system for determining text coherence |
CN101228577B (en) * | 2004-01-12 | 2011-11-23 | 语音信号技术公司 | Automatic speech recognition channel normalization method and system |
US20080208578A1 (en) * | 2004-09-23 | 2008-08-28 | Koninklijke Philips Electronics, N.V. | Robust Speaker-Dependent Speech Recognition System |
US8175877B2 (en) * | 2005-02-02 | 2012-05-08 | At&T Intellectual Property Ii, L.P. | Method and apparatus for predicting word accuracy in automatic speech recognition systems |
KR100714721B1 (en) | 2005-02-04 | 2007-05-04 | 삼성전자주식회사 | Method and apparatus for detecting voice region |
US8202098B2 (en) * | 2005-02-28 | 2012-06-19 | Educational Testing Service | Method of model scaling for an automated essay scoring system |
DE102005030855A1 (en) * | 2005-07-01 | 2007-01-11 | Müller-BBM GmbH | Electro-acoustic method |
US7725316B2 (en) * | 2006-07-05 | 2010-05-25 | General Motors Llc | Applying speech recognition adaptation in an automated speech recognition system of a telematics-equipped vehicle |
CN101154380B (en) * | 2006-09-29 | 2011-01-26 | 株式会社东芝 | Method and device for registration and validation of speaker's authentication |
JP2008298844A (en) * | 2007-05-29 | 2008-12-11 | Advanced Telecommunication Research Institute International | Noise suppressing device, computer program, and speech recognition system |
US20100094622A1 (en) * | 2008-10-10 | 2010-04-15 | Nexidia Inc. | Feature normalization for speech and audio processing |
US20130158996A1 (en) * | 2011-12-19 | 2013-06-20 | Spansion Llc | Acoustic Processing Unit |
US9953646B2 (en) | 2014-09-02 | 2018-04-24 | Belleau Technologies | Method and system for dynamic speech recognition and tracking of prewritten script |
US9824684B2 (en) | 2014-11-13 | 2017-11-21 | Microsoft Technology Licensing, Llc | Prediction-based sequence recognition |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0301199A1 (en) * | 1987-07-09 | 1989-02-01 | International Business Machines Corporation | Normalization of speech by adaptive labelling |
EP0487309A2 (en) * | 1990-11-20 | 1992-05-27 | Canon Kabushiki Kaisha | Pattern recognition method and apparatus using a neural network |
US5185848A (en) * | 1988-12-14 | 1993-02-09 | Hitachi, Ltd. | Noise reduction system using neural network |
-
1994
- 1994-07-29 US US08/283,271 patent/US5604839A/en not_active Expired - Lifetime
-
1995
- 1995-07-26 EP EP95111784A patent/EP0694906B1/en not_active Expired - Lifetime
- 1995-07-26 DE DE69518705T patent/DE69518705T2/en not_active Expired - Lifetime
- 1995-07-27 JP JP7222471A patent/JPH08110793A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0301199A1 (en) * | 1987-07-09 | 1989-02-01 | International Business Machines Corporation | Normalization of speech by adaptive labelling |
US5185848A (en) * | 1988-12-14 | 1993-02-09 | Hitachi, Ltd. | Noise reduction system using neural network |
EP0487309A2 (en) * | 1990-11-20 | 1992-05-27 | Canon Kabushiki Kaisha | Pattern recognition method and apparatus using a neural network |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0871157A3 (en) * | 1997-04-11 | 1999-06-16 | Nokia Mobile Phones Ltd. | A method and a device for recognising speech |
EP0871157A2 (en) * | 1997-04-11 | 1998-10-14 | Nokia Mobile Phones Ltd. | A method and a device for recognising speech |
US6772117B1 (en) | 1997-04-11 | 2004-08-03 | Nokia Mobile Phones Limited | Method and a device for recognizing speech |
EP1113419A1 (en) * | 1999-12-28 | 2001-07-04 | Sony Corporation | Model adaptive apparatus and model adaptive method, recording medium, and pattern recognition apparatus |
US7043425B2 (en) | 1999-12-28 | 2006-05-09 | Sony Corporation | Model adaptive apparatus and model adaptive method, recording medium, and pattern recognition apparatus |
US6920421B2 (en) | 1999-12-28 | 2005-07-19 | Sony Corporation | Model adaptive apparatus for performing adaptation of a model used in pattern recognition considering recentness of a received pattern data |
US7003455B1 (en) | 2000-10-16 | 2006-02-21 | Microsoft Corporation | Method of noise reduction using correction and scaling vectors with partitioning of the acoustic space in the domain of noisy speech |
EP1199712A2 (en) * | 2000-10-16 | 2002-04-24 | Microsoft Corporation | Noise reduction method |
EP1199712A3 (en) * | 2000-10-16 | 2003-09-10 | Microsoft Corporation | Noise reduction method |
US7254536B2 (en) | 2000-10-16 | 2007-08-07 | Microsoft Corporation | Method of noise reduction using correction and scaling vectors with partitioning of the acoustic space in the domain of noisy speech |
EP1246165A1 (en) * | 2001-03-28 | 2002-10-02 | Matsushita Electric Industrial Co., Ltd. | Keyword detection in a noisy signal |
US6985859B2 (en) | 2001-03-28 | 2006-01-10 | Matsushita Electric Industrial Co., Ltd. | Robust word-spotting system using an intelligibility criterion for reliable keyword detection under adverse and unknown noisy environments |
EP1326233A2 (en) * | 2001-12-28 | 2003-07-09 | Kabushiki Kaisha Toshiba | Apparatus and method for speech recognition in noise |
EP1326233A3 (en) * | 2001-12-28 | 2004-07-28 | Kabushiki Kaisha Toshiba | Apparatus and method for speech recognition in noise |
US7117148B2 (en) | 2002-04-05 | 2006-10-03 | Microsoft Corporation | Method of noise reduction using correction vectors based on dynamic aspects of speech and noise normalization |
US7181390B2 (en) | 2002-04-05 | 2007-02-20 | Microsoft Corporation | Noise reduction using correction vectors based on dynamic aspects of speech and noise normalization |
US7542900B2 (en) | 2002-04-05 | 2009-06-02 | Microsoft Corporation | Noise reduction using correction vectors based on dynamic aspects of speech and noise normalization |
WO2005036525A1 (en) * | 2003-10-08 | 2005-04-21 | Philips Intellectual Property & Standards Gmbh | Adaptation of environment mismatch for speech recognition systems |
GB2422237A (en) * | 2004-12-21 | 2006-07-19 | Fluency Voice Technology Ltd | Dynamic coefficients determined from temporally adjacent speech frames |
CN111489754A (en) * | 2019-01-28 | 2020-08-04 | 国家电网有限公司客户服务中心 | Telephone traffic data analysis method based on intelligent voice technology |
Also Published As
Publication number | Publication date |
---|---|
DE69518705D1 (en) | 2000-10-12 |
EP0694906B1 (en) | 2000-09-06 |
DE69518705T2 (en) | 2001-01-04 |
JPH08110793A (en) | 1996-04-30 |
US5604839A (en) | 1997-02-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP0694906A1 (en) | Method and system for speech recognition | |
CA2147772C (en) | Method of and apparatus for signal recognition that compensates for mismatching | |
US5148489A (en) | Method for spectral estimation to improve noise robustness for speech recognition | |
US4905286A (en) | Noise compensation in speech recognition | |
EP0966736B1 (en) | Method for discriminative training of speech recognition models | |
US5806029A (en) | Signal conditioned minimum error rate training for continuous speech recognition | |
US5023912A (en) | Pattern recognition system using posterior probabilities | |
US6493667B1 (en) | Enhanced likelihood computation using regression in a speech recognition system | |
US7269556B2 (en) | Pattern recognition | |
EP0470245B1 (en) | Method for spectral estimation to improve noise robustness for speech recognition | |
US6421640B1 (en) | Speech recognition method using confidence measure evaluation | |
Bahl et al. | Multonic Markov word models for large vocabulary continuous speech recognition | |
US5684924A (en) | User adaptable speech recognition system | |
US6202047B1 (en) | Method and apparatus for speech recognition using second order statistics and linear estimation of cepstral coefficients | |
US6725196B2 (en) | Pattern matching method and apparatus | |
WO1997010587A9 (en) | Signal conditioned minimum error rate training for continuous speech recognition | |
JPH05257492A (en) | Voice recognizing system | |
US5345535A (en) | Speech analysis method and apparatus | |
Welling et al. | A model for efficient formant estimation | |
EP1074018B1 (en) | Speech recognition system and method | |
EP0435336A2 (en) | Reference pattern learning system | |
Rose et al. | Robust speaker identification in noisy environments using noise adaptive speaker models | |
EP0309561B1 (en) | An adaptive threshold voiced detector | |
KR0170317B1 (en) | Voice recognition method using hidden markov model having distortion density of observation vector | |
Pan et al. | An efficient vector-quantization preprocessor for speaker independent isolated word recognition |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): DE FR GB |
|
17P | Request for examination filed |
Effective date: 19960725 |
|
17Q | First examination report despatched |
Effective date: 19990519 |
|
GRAG | Despatch of communication of intention to grant |
Free format text: ORIGINAL CODE: EPIDOS AGRA |
|
GRAG | Despatch of communication of intention to grant |
Free format text: ORIGINAL CODE: EPIDOS AGRA |
|
GRAH | Despatch of communication of intention to grant a patent |
Free format text: ORIGINAL CODE: EPIDOS IGRA |
|
RIC1 | Information provided on ipc code assigned before grant |
Free format text: 7G 10L 15/20 A |
|
GRAH | Despatch of communication of intention to grant a patent |
Free format text: ORIGINAL CODE: EPIDOS IGRA |
|
GRAA | (expected) grant |
Free format text: ORIGINAL CODE: 0009210 |
|
AK | Designated contracting states |
Kind code of ref document: B1 Designated state(s): DE FR GB |
|
REF | Corresponds to: |
Ref document number: 69518705 Country of ref document: DE Date of ref document: 20001012 |
|
ET | Fr: translation filed | ||
PLBE | No opposition filed within time limit |
Free format text: ORIGINAL CODE: 0009261 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT |
|
26N | No opposition filed | ||
REG | Reference to a national code |
Ref country code: GB Ref legal event code: IF02 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: GB Payment date: 20130624 Year of fee payment: 19 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: DE Payment date: 20130731 Year of fee payment: 19 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: FR Payment date: 20130712 Year of fee payment: 19 |
|
REG | Reference to a national code |
Ref country code: DE Ref legal event code: R082 Ref document number: 69518705 Country of ref document: DE Representative=s name: GRUENECKER, KINKELDEY, STOCKMAIR & SCHWANHAEUS, DE |
|
REG | Reference to a national code |
Ref country code: DE Ref legal event code: R119 Ref document number: 69518705 Country of ref document: DE |
|
REG | Reference to a national code |
Ref country code: GB Ref legal event code: 732E Free format text: REGISTERED BETWEEN 20150108 AND 20150114 |
|
REG | Reference to a national code |
Ref country code: DE Ref legal event code: R082 Ref document number: 69518705 Country of ref document: DE Representative=s name: GRUENECKER PATENT- UND RECHTSANWAELTE PARTG MB, DE Effective date: 20150126 Ref country code: DE Ref legal event code: R081 Ref document number: 69518705 Country of ref document: DE Owner name: MICROSOFT TECHNOLOGY LICENSING, LLC, REDMOND, US Free format text: FORMER OWNER: MICROSOFT CORP., REDMOND, WASH., US Effective date: 20150126 |
|
GBPC | Gb: european patent ceased through non-payment of renewal fee |
Effective date: 20140726 |
|
REG | Reference to a national code |
Ref country code: FR Ref legal event code: ST Effective date: 20150331 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: DE Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20150203 |
|
REG | Reference to a national code |
Ref country code: DE Ref legal event code: R119 Ref document number: 69518705 Country of ref document: DE Effective date: 20150203 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: FR Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20140731 Ref country code: GB Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20140726 |